All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
DartNode $13.99 Flash Deal: The 9Gbps Houston Firehose vs. The 5 IOPS Soda Straw
We managed to snag one of the 10 Flash Deal units from the 2025 Black Friday / Cyber Monday megathread.
After nearly three months of ownership, we’ve realized this isn't just a VPS—it’s a performance art piece dedicated to the sheer absurdity of budget hosting.
The Legend of Daniel
First off, we haven't canceled this out of pure respect for Daniel.
Rumor has it he's pulling 14-hour shifts racking hardware at TRG Houston One.
Because of our ongoing tickets regarding the shared IPv6 /64 and the VNC "FIN" packet issue, the team agreed to not start our 1-year billing clock until things are fixed.
Effectively, we are hosting a 10Gbps static site for free while the billing clock stays frozen in time.
The "Blink" Metric
Forget IOPS. we're introducing a new Houston-standard benchmark: The Blink.
- Standard Login: 13 cursor blinks before a prompt appears.
- Optimized Login (
--norc): 1-4 blinks. - The VNC Experience: A perfect 10-minute countdown until the server sends a TCP FIN packet and ghosts you. It’s not a bug; it’s a "focus timer."
The Benchmark Roast
We ran a full suite of fio and networking tests.
The results are a digital coma:
- YABS: ✔ Online (/32) / ✔ Online (/128), IPv6 available yeah!
- nws.sh: near-line-rate 10Gbps awesomeness. We hit 9.2 Gbps peak to Miami. The network firehose is so wide that it can deliver a 40GB Blu-ray in 35 seconds—shame the disk needs 57 minutes to write it.
- latency-check: A 9,029ms maximum clat on random reads. It doesn't lag; it goes to the afterlife and back.
- mixed-workload: The 6 Gbps network is ~2000x faster than the 3 Mbps disk.
- max-throughput: Big reads hit 3.5 GB/s (Ferrari engine), but random 4k hits 1.8 MB/s (square wheels).
- sustained-write: 5 IOPS. 2001 called; they want their USB 1.1 flash drive back.
The "Art Piece" Live Gallery
Since the disk had a literal 133-second stroke during a 10GB write test, we've moved the entire site to a RAM buffer served by Caddy.
It now loads at the full 9.2 Gbps peak capacity of the Snaju network, completely bypassing the Houston rust.
View our gallery: https://freaky-fast-digital-coma.yoursunny.dev
Disclaimer:
Announcement and poll options were written by Google AI based on real benchmark results, which can be access from the above website.
This page is a tribute to the hustle and grind, not a complaint about the hardware.
No Daniel was harmed in making this art piece.
- If you had to choose between these "features," which defines the $13.99 DartNode Art Piece?26 votes
- The Freaky Fast Firehose (I live for the 9 Gbps bursts and Daniel's hustle)23.08%
- The Digital Coma (I enjoy the 133-second stalls and 13-blink logins)26.92%
- The Frozen Clock (I just want to see how long I can host a static site for free)  7.69%
- Daniel's Mental Fortitude (I'm only here to support the man racking the servers)42.31%

Comments
it makes me deeply sad to see you use ai for stuff like this
i always thought you were funny without it
On this DartNode, we have invested:
We don't have any push-ups left for the write-up, hence we delegated the task to our digital assistant.
So RAM is fast my dude! Buy more RAM, like 64GB, create RAM-disk only server.
Everything fast.
We can't afford to buy more RAM, but there's always https://downloadmoreram.com/ .
Downloading RAM at 9Gbps is definitely faster than mounting the
/swapfileat 5 IOPS.Remember the day VirBot sold 12GB RAM with 5GB NVMe?
Update: the "art gallery" now offers realtime
vmstatdiagrams showing CPU iowait and steal values.https://freaky-fast-digital-coma.yoursunny.dev/
Screenshot:

Source code:
https://github.com/yoursunny/freaky-fast-digital-coma
Artwork currently offline because VPS cannot power on.
When Daniel wakes up he'll press the power buttons, we hope.
Meanwhile, we did 9 push-ups in sync with the 9Gbps firehose.
lol this is still going on?
The SSH connection is currently 3 blinks, ending with "no route to host".
Will you do 3 push-ups for Daniel?
Artwork is back online after 15 hours of outage (since we added UptimeRobot when we posted the offline notice).
Digital coma continues it's 23 IOPS.
https://serververify.com/benchmarks/c5f04510-1e62-4817-84d8-51f6992875f2
This was just in:
[DartNode Network Alert] Case #20 SSD-2 Drive Failure
Hardware Issue Notification — SSD-2 Drive Failure
We are notifying you of a hardware issue that may impact your services at DartNode. Our team is working to address this as quickly as possible.
Started At
2/27/2026 21:00 UTC
Expected Resolution
2/28/2026 00:00 UTC
Details
Our team has identified that a drive (/dev/sdj) in the RAID md0 array is in a failure state. To prevent further issues and restore the system to full health, we are establishing an emergency maintenance window, effective immediately.
Please be advised that customers may experience temporary downtime while we replace the drive and synchronize the array. We will provide another status update as soon as the restoration is complete.
Thanks,
DartNode NOC Team
The artwork was offline for about an hour, right after this notice, but has since recovered.
We peeked vmstat for a minute, on a mostly idle server, and saw iowait shoot up to 16 briefly.
Boom. Exactly.
You succeeded in drawing sunny out from behind the cloud, @fluffernutter
They may be enabling QEMU's emulated hardware watchdog for all customers soon, which should help somewhat with reliability, given the VMs don't like to reboot on their own without manual intervention.
Update: our billing clock thawed on 2026-03-01, followed the resolution of VNC errors.
There seems to be a significant congestion today.
SSH will not connect, regardless of how many blinks we wait.
Yo-sunny, long time no see
In Houston we have a problem.
IOPS increased somewhat but still far from NVMe speeds.
10Gbps fire hose has been choked.